#asahi · 2024-05-04 10:02
PaulFertsernano: probabl you can just do "diskutil apfs resizeContainer" ?#asahi · 2024-05-04 10:00
PaulFertsernano: huh, the instructions say they result in deleting all the containers.#asahi · 2024-05-04 09:56
PaulFertsernano: probably you can just follow https://discussions.apple.com/docs/DOC-250007691 instead#asahi · 2024-05-04 09:53
PaulFertsernano: let me ask for what fraction of the retail price are you selling it?#asahi · 2024-05-04 09:51
PaulFertsernano: so https://support.apple.com/en-us/108900 but with idevicerestore should work. I think the battery cycle counter won't be reset either way.#asahi · 2024-05-04 09:48
PaulFertsernano: I think you can repartition the main storage from recovery too but since you want absolutely no traces of Asahi I'm not sure if that's going to guarantee it.#asahi · 2024-05-04 09:47
PaulFertserYep#asahi · 2024-05-04 09:47
PaulFertserusbmuxd might also be needed to be latest, not sure.#asahi · 2024-05-04 09:47
PaulFertserA USB-C cable, yes.#asahi · 2024-05-04 09:47
PaulFertserAnd then just idevicerestore --erase --latest#asahi · 2024-05-04 09:46
PaulFertserYou might need to build the app and its lib from latest git sources yourself, do not count on distro-packaged version.#asahi · 2024-05-04 09:45
PaulFertseridevicerestore should work on any GNU/Linux system so another mac isn't needed.#asahi · 2024-05-04 09:45
PaulFertser(or with another macOS device and proprietary tool)#asahi · 2024-05-04 09:44
PaulFertsernano: probably DFU with idevicerestore for making it super vanilla clean?#asahi · 2024-05-03 12:53
PaulFertserI hope real devs can give you a better suggestion here.#asahi · 2024-05-03 12:52
PaulFertserfrek: what I'd try would be to use https://www.man7.org/linux/man-pages/man1/perf-sched.1.html to perf sched record for sound playing normally (when you do not move mouse cursor on ext display) and with stuttering (when you do) and then showing results in a bug report on https://github.com/displaylink/evdi#asahi · 2024-05-03 12:42
PaulFertserFor the reference, DP chaining also requires MST because for chaining the display itself includes an MST "switch" with two downstream ports (one internal for the display itself, one external to chain to the next).#asahi · 2024-05-03 12:35
PaulFertserbluetail: DisplayLink is like a USB-attached video card.#asahi · 2024-05-03 12:34
PaulFertserYes, yes. I mean there still is some advantage of the newer DisplayLink devices for certain usecases with macs.#asahi · 2024-05-03 12:33
PaulFertserSo the kernel part is GPLv2, the rest is MIT, looks like proper Free Software to me.#asahi · 2024-05-03 12:31
PaulFertserbluetail: yes, and Apple doesn't have hardware to allow DP chaining so you're limited to just one device. While with DisplayLink USB SS you can have few external displays.#asahi · 2024-05-03 12:30
PaulFertserj`ey: for testing indeed, I'm suggesting to do that to tell if the issue affects alsa or just the sound server.#asahi · 2024-05-03 12:30
PaulFertserbluetail: really? I thought the evdi kernel driver is fully proper GPLv2?#asahi · 2024-05-03 12:29
PaulFertserj`ey: how is it not supported? Yes, you get no sound profile for smart DSP but so what?#asahi · 2024-05-03 12:28
PaulFertserTry to play sound directly via alsa is a user level thing.#asahi · 2024-05-03 12:28
PaulFertserfrek: I am not suggesting you start chasing latency bottlenecks in evdi kernel code :)#asahi · 2024-05-03 12:27
PaulFertserfrek: alt dp isn't really a replacement because it only supports one external display. While with displaylink you have few more.#asahi · 2024-05-03 12:25
PaulFertserfrek: probably I have just too little clue about pipewire. If I had such an issue I'd stop pipewire and would start testing with alsa directly to isolate the problematic part.#asahi · 2024-05-03 12:22
PaulFertserRight#asahi · 2024-05-03 12:22
PaulFertserfrek: damn :/ unfortunately I have no more guesses. My current theory is that evdi doesn't let latency requirements of your sound server to be met because it affects bluetooth and integrated sound card all alike.#asahi · 2024-05-03 12:17
PaulFertserfrek: just 100 and 2000 should be enough.#asahi · 2024-05-03 12:17
PaulFertserGood thing we have _online manuals_ available just one "man anything" command away.#asahi · 2024-05-03 12:14
PaulFertserfrek: right, and do you get no difference at all no matter what --pipewire-buffer value you use?#asahi · 2024-05-03 12:11
PaulFertserfrek: and what --pipewire-buffer have you tried? And do you see in mpv output that it's using pipewire directly?#asahi · 2024-05-03 12:10
PaulFertserfrek: playing audio _with mpv_ only#asahi · 2024-05-03 12:08
PaulFertserfrek: and also mpv allows to easily control that buffering we're interested in testing to find possible workarounds for your issue.#asahi · 2024-05-03 12:06
PaulFertserfrek: mpv shows videos from youtube nicely#asahi · 2024-05-03 11:56
PaulFertserfrek: try with mpv#asahi · 2024-05-03 11:56
PaulFertserfrek: it's supposed to be set before a playback app, but some apps will overwrite anyway.#asahi · 2024-05-03 11:55
PaulFertserfrek: that env variable doesn't affect systemd units obviously :)#asahi · 2024-05-03 11:55
PaulFertserfrek: you can also try playing audio with plain mpv and different values for --pipewire-buffer (make sure it uses pipewire ao, not pulseaudio or alsa).#asahi · 2024-05-03 11:51
PaulFertserThis looks quite demanding: The default maximum latency was reduced from 170ms to 42ms. https://gitlab.freedesktop.org/pipewire/pipewire/-/releases/0.3.44#asahi · 2024-05-03 11:47
PaulFertserBut in general my idea is to increase buffering (and hence decrease requirements for low latency scheduling). And it seems to be somewhere in the sound daemon you use, the one that mixes the sounds from different apps.#asahi · 2024-05-03 11:46
PaulFertserI'm not sure what exactly sound stack you're using and how, what's the actual path the sound goes through userspace before enters the kernel. It depends on your distro and settings.#asahi · 2024-05-03 11:40
PaulFertserfrek: so try asking pipewire to use more buffering the way I suggested#asahi · 2024-05-03 11:15
PaulFertserOr you can do this to change defaults temporarily: pw-metadata -n settings 0 clock.force-rate <samplerate>; w-metadata -n settings 0 clock.force-quantum <buffersize>#asahi · 2024-05-03 11:13
PaulFertserYou can try setting that env variable prior to running an app that plays music, it's per-stream/per-app essentially.#asahi · 2024-05-03 11:12
PaulFertserfrek: I suggest you try running pipewire with PIPEWIRE_LATENCY=200/1000 to see if it makes any difference. I do not know how exactly to do that on your distro but it should be easy.#asahi · 2024-05-03 11:08
PaulFertserfrek: do you have any underflows mentioned in "journalctl -u pipewire-pulse.service" ?#asahi · 2024-05-03 11:06
PaulFertserOr pulseaudio?